Poster #92 - Developing a scale of cognitive stimulation using the Home Observation Measurement of the Environment- Infant/Toddler (HOME-IT)

• In Event: Poster Session 09

Abstract

Background: Early parenting behaviors that promote cognitive stimulation (CS) are positively associated with cognitive (LeWinn et al., 2020) and academic outcomes (Herbers et al., 2011), and executive function skills (Rosen et al., 2020). Existing CS measures are lengthy and rely heavily on live, in-home parent-child interaction observations, making them impractical in large, geographically dispersed samples (Rasheed & Yousafzai, 2015; Reese et al., 2015). Thus, there is a need for brief measures of CS to ease the burden of data collection and to build our understanding of this modifiable protective factor in early childhood. The primary objective of this study was to develop a brief measure of CS that could be adapted for remote assessments and repeat administration in large epidemiological samples. Methods: We leveraged data from three NIH Environmental Influences on Child Health Outcomes Program (ECHO) cohorts that collected the HOME-Infant-Toddler (HOME-IT age 0-3 years) version to examine the reliability of a brief scale of CS in a diverse epidemiological sample. The current analysis is based on 2,518 participants with available data on the HOME-IT. The HOME-IT was completed at around 2 years of age (mean=1.51, SD = 0.51); 48% of children were female, and 5% of children and 3% of mothers identified as Hispanic. The average maternal age at the time of the HOME-IT was 26.0 years (SD = 5.87), and maternal education ranged between high school or less than a high school education (3%) to a bachelor’s degree or above (17%). The analysis occurred in three stages using a statistical analysis plan for evaluating item response theory (IRT) modeling assumptions (unidimensionality and local independence), IRT model fit, monotonicity, scalability, item fit, and differential item functioning (DIF). Results: After a series of analyses, the final bifactor CFA consisted of 18 items across four local factors with acceptable fit (CFI=0.97; RMSEA=0.02). All items had r-square >0.3, and only two item pairs had a residual correlation >0.2. The variance explained by the general factor was greater than that explained by local factors, supporting sufficient unidimensionality of these items. IRT analysis was then conducted on these 18 items. Six items had a fit test ratio >0.3; 2 items had a slope parameter <1; and 4 items showed DIF on at least two comparisons. Content experts (C.O. and B.R.C.) reviewed items along with the above results and selected 10 items to be included in the final CS scale. Discussion: Results indicate the feasibility of developing a reliable 10-item cognitive stimulation scale. Strengths of this research include applying best practices for scale development in a large and more diverse sample than in previous research on CS scales. The next steps in this research include testing this brief 10-item scale against a gold standard measure to assess precision and determine an appropriate cut-off score to identify families that would benefit from additional interventions and support.

Group Authors

ECHO Program

Authors

• Cindy Ola Trevino, University of Washington
  Presenting Author

• Kaja Z. LeWinn, University of California - San Francisco

• Sara Nozadi, University of New Mexico

• Adaeze C Wosu, Johns Hopkins University

• Catherine J. Karr, University of Washington

• Xiaodan Tang, Northwestern University

• Jin-Shei Lai

• Brent Russell Collett, University of Washington